Automatic Tracking of Muscle Fiber Direction in Ultrasound Images Based on Improved Kalman Filter

Round 1

Reviewer 1 Report

This paper presents the tracking of muscle fiber direction in ultrasound images based on improved Kalman Filter. Its a nice paper and it can be improved. The authors Applied their research ısing NVIDIA 3070. The paper can be improved by adding more images from the dataset rather than single image and deep learning can be discussed in more details by giving the structure. The results also needs discussion.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript is quite interesting on the application of an improved Kalman filter. On this respect, can you discuss if other kalman filters that are recently aplied, are suitable for your research? See for example the Kalman filter version applied in the reference: Kalman observers in estimating the states of chaotic neurons for image encryption under MQTT for IoT protocol

The European Physical Journal Special Topics, 2021.

As you show in Table 2. The root mean square error(RMSE) of the results obtained by the five methods, the Kalman filters that you apply are KF, EKF, LSTM, in the reference recomended above, you can see the unscented Kalman filter, please, discuss on the possibility of be useful in your work.

As you mention in the abstract in the following sentences: Firstly, the measurement value of the muscle fiber direction is obtained by 6

introducing a reference line into the ultrasound muscle image based on deep learning. Secondly, the 7

framework of Kalman filter is improved by introducing a set of neural units. Finally, the optimal 8

estimated value of muscle fiber direction is obtained by combining the measured value with the 9

improved Kalman filter…. You use the improved Kalman filter just in one the the three stages of your method. On these issues: Are the deep learning techniques and the neural units, the most suibale ones for your work? WIll you explore more options on these methods to improve your work and still using the improved Kalman filter or the unscented kalman filter?

At the end of Section 1, which is very long and may be shortened, you write part in the sentences: The remaining content is organized as follows. In the second part, we introduce the 115

method we used. In the third part, we show the experiment of extracting and tracking the 116

direction of muscle fibers in the sequence of ultrasound images. And then the experimental 117

results will been discussed in the fourth part. Our final summary of the paper is in the last 118

part…. You may change the Word part by the Word Section, as done in all research articles in all journals.

In Section 2. Materials and Methods 120

Deep learning is applied to many areas of images. In the previous work, a reference 121

line is introduced into the image of the muscle fiber, and the ResNet-50 deep learning 122

network[27] is used to judge the relationship between the muscle fibers and the line through 123

the sub-images…. When you mention “In the previous work”, you must cite such a work, otherwise the reader cannot imagine what are you trying to discuss. May be reference [27] is such a previous work, and then must be cited in that sentence, as for example in this manner … In the previous work reported in [27]…

In line 138 you mention: Firstly, the Kalman filter uses the process model of the system to predict the next state of the system… again, you must indicate if the process model is equation (1), which is mentioned in the sentence: Using formula (1), assuming that the current system state is k , according to

the system model, the current state can be predicted based on the previous state….In this respect, you may consider revising the recommended reference mentioned above: Kalman observers in estimating the states of chaotic neurons for image encryption under MQTT for IoT protocol… for the prediction of states.

When you give Figure 2. The structure of the combination of NARX model and Kalman Filter algorithm.,.. the NARX is quite used in several recent Works, and you may discuss on the possibility of implementing it in a portable equipment, as in a field-programmable gate array, as done in the reference: Pipeline FPGA-based Implementations of ANNs for the Prediction of up to 600-steps-ahead of Chaotic Time Series

Journal of Circuits, Systems and Computers 30 (09), 2150164, 2021. .. This manuscript also applies RMSE and other eqaution for the error measument, as you mnetion in your manuscript in line 214: We use the root mean square error to reflect accuracy.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper presents an interesting idea and topic. The following is the reviewer's suggestion to the paper:
1. Recently, a topic related to the medical is potential for research. This paper presents a topic related to medical science and human health condition as mentioned in lines 3-4. I suggest the Authors include a Co-Author in the related field As I noted that all Authors are from Computer Science and Engineering only; or the Authors could add the related specialist in the Acknowledgement to review the result of the proposed method in the field of medical science.
2. Please describe in more detail how the data is collected.
3. This paper presents the application of Kalman Filter (KF) and Extended Kalman Filter (EKF). When researchers used these methods, they usually compare with Particle Filter (PF). Can the Authors provide the comparison of KF, EKP, and PF in the revised paper as well?

The Authors could see the following suggestion literature:
- An overview of existing methods and recent advances in sequential Monte Carlo
- Machine condition prognosis based on sequential Monte Carlo method

4. Apart from the KF method, the Authors used NARX, is there any specific reason why this method is selected instead of the ARMA, ARIMA, or GARCH model?
5. The legend of Figure 4 is not readable.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors applied the reviewer comments accordigly.

Reviewer 2 Report

The updated version of this manuscript has been greately improved. it is a very nice work and it must be accepted as it is.

Reviewer 3 Report

Dear Authors,

Thank you for providing a revise version. I have checked the resubmitted documents and I have no further comments.

Kind regards,

- Reviewer -